972 NATURE June 17, 1950 Vol.

165

Application of the Kozeny Equation to derived from an electrical measurement and k 0

Consolidated Porous Media
RECENTLY, Carman1 discussed the applicability of
the Kozeny equation to the determination of the
specific surface area of consolidated porous media
and concluded that it should be applicable if certain
specified conditions were fulfilled. Carman apparently
envisaged the use of the Kozeny equation with a
constant, k, of 5 ·O, which is the constant found
applicable to the case of fluid flow through uncon-
solidated porous media. Carman has shown• that k
may be expressed as the product of two parameters,
namely, a shape factor, k 0 , and a term (Le/L)• which
we shall call tortuosity, T. He has adduced experi-
mental and theoretical evidence to show that the
variation of k 0 with pore geometry is likely to be
within the approximate range 2·0-2·5 for most porous
media, and this conclusion appears to have received
general confirmation. For unconsolidated porous
media in the porosity range 0·25-0·90, Carman has
suggested that k 0 is about 2 ·5, and it is thus implicit
that within this range T has a rather constant value
of 5·0/2·5 = 2·0.
Table 1
• F
(Carman)
F
(Fricke and Morse)
(Archie)
0·25 5·7 5·5
0·30 4·7 4·5
0·35 4·0 3·8
0·40 3·5 3·3
0·45 3·1 2·8
We suggest that T can be derived from the ratio
of the resistivity of a porous medium saturated with
a conducting fluid to the resistivity of the fluid. This
ratio has been shown by Archie• to be a constant for
any porous medium and was given the name forma-
tion factor, F, by him. Archie showed empirically
that for unconsolidated porous media F = e-1 ·•. We
have shown that F = T 1 l 2 / e, and details of the de-
rivation are being published elsewhere.
Table I gives a comparison of values of F com-
puted from (2 ·0)11•/ e (Carman), (3 - e)/2e (Fricke and
Morse' for dispersed non-conducting spheres) and
e-1 • 3 (Archie).
assumed, a value of k which may greatly exceed 5 ·0
may be computed. This k is then used in the Kozeny
equation, Sp = (e/kK)1I•, where K is the coefficient
of permeability, to compute Sp, specific surface per
unit of pore volume. The accuracy of such computa-
tions is difficult to check since there is no suitable
independent method for measuring the specific sur-
face area of consolidated media ; the gas adsorption
method does not necessarily measure the same surface
area as that determined in flow experiments. Carman
has himself shown•, however, that in unconsolidated
porous media the expression Pn/cr appears to give
a close approximation to the specific surface area
per unit of pore volume. PD is the pressure that must
be applied to a non-wetting phase in order to initiate
displacement of a wetting fluid from a saturated
porous medium, and cr is the interfacial tension
between the two phases.
We have investigated the agreement between sur-
face areas of consolidated porous media based on
values of k = e•F•k 0 with those computed from the
relationship Pn/cr.
Table 2 shows a few results of measurements we
have made, and values of Sp which have been cal-
culated from them. A value of k 0 = 2·5 was assumed.
F
The measurement of Pn can only be made with
accuracy if the pore-size distribution of the media
6·1 used is reasonably uniform. To ensure this, measure-
4·8
3·9 ments have been made on consolidated porous media
3·3 fabricated by sintering 'Pyrex' and alundum grains
2·8
of rather uniform particle sizes (supplied by Corning
Glass Works, New York, and the Norton Company,
Worcester, Massachusetts, respectively). However,
measurements on certain natural consolidated sand-
stones, the results from two being quoted here, have
also been carried out and have been found very
reproducible. The fact that consolidated porous
media may be prepared with smooth faces materially
assists the accuracy with which an observation of
PD can be made.
It will be observed that the agreement between
the surface areas determined by the two methods
is generally satisfactory. The calculation of values
of Sp based on values of T derived from measured
values of F thus appears promising.
Table 2

Material K
(cm.• x 10-11 ) • F I
I
k (dynes/cm.•
PD
x 10•)
I o
(dynes/cm.)
I Sp=S»Ja
(cm. 1/cm.') I
Sp= (2•6 • K)- 111Jr1
(cm. 1/cm.')
Alundum 680 0·258 13·7 31·2 5·17 43·0 1,200 1,100
Alundum 660 0·254 11·3 20·5 7·24 43·0 1,680 1,370
Alundum 63 0·236 11·5 18·4 29•2 54·2 5,400 4,500
Pyrex 8·1 0·374 4·4 6·8 124·0 54•2 22,900 26,100
Pyrex 3900 0·286 6·2 7·8 5·65 54·2 1,030 870
Nicholls Buff SS 230 0•200 12·5 15·6 11·9 48•0 2,760 2,360
Berea 88 890 0·225 11·7 17·3 6·02 43·0 1,400 1,200

It will be observed that the comparison is reason- Grateful acknowledgment of permission to publish
ably good. One reason for the discrepancy between this note is given to Dr. Paul D. Foote, executive
Archie's values of F and those computed from the vice-president, Gulf Research and Development
Carman tortuosity seems certainly to lie in the fact Company, Harmarville, Pennsylvania.
that no allowance for a decrease in tortuosity with M. R. J. WYLLIE
increase in porosity is made in the latter formulation ; WALTER D. ROSE
that is, we believe that the relative constancy of k Gulf Research and Development Company,
observed by Carman in the case of unconsolidated Harmarville, Pa. Jan. 23.
porous media is the fortuitous consequence of com- 1
Carman, P. C., Farad. Soc. Discussions, "Interaction of Water and
pensating changes in k 0 and T with change in e. Porous Materials", 72 (1948).
However, if k 0 is relatively constant over a wide 'Carman, P. C., Trans. Inst. Chem. Eng., 15, 150 (1937).
'Archie, G. E., Trans • .Amer. Inst. Mech. Eng. (Petroleum Division),
range of geometrical pore shapes-and the recent 146, 54 (1942). .
results of Coulson 5 support this view-k can be cal- • Fricke, H., and Morse, S., Ph'V•- Rev., 25, 361 (1925).
• Coulson, J.M., Trans. Inst. Chem. Eng. (in the press).
culated if T can be separately determined. If T is 'Cannan, P. C., Soil Sci., 52, 1 (1941).

© 1950 Nature Publishing Group